Measuring valence-dependent belief updating may allow for the prediction of mood episodes among patients with bipolar disorder, according to the findings from an observational prospective study published in eLife.

Patients (N=36) who were treated at the Psychiatric University Clinic of Parma in Italy were recruited for this study between 2013 and 2014. All participants had a diagnosis of bipolar disorder I or II and did not fit the criteria for Rapid Cycling. Patients had psychiatric hospital visits every 2 months for 5 years in which the patients were asked to perform a behavioral task.

During the task, 40 descriptions of negative life events were presented. Patients had to estimate the likelihood that they would experience the event in the future. The patients were informed of the actual probability of each event for someone with their demographics and asked to reevaluate the likelihood for each event. Following the tasks, patients were assessed by the Beck Depression Inventory (BDI-II) and the Revised Life Orientation Test (LOT-R).

Continue Reading

The investigators calculated the bias for each patient by comparing their preliminary and updated probabilities of each life event. This valence-dependent belief updating score indicated whether patients had a more optimistic bias, indicating they originally believed they had a high likelihood of experiencing a bad life event or a pessimistic update bias, indicating they underestimated their likelihood of the negative event.

A larger positive bias predicted a greater future time in euthymia (b, 0.589; 95% CI, 0.122-1.129; P =.004) as did preliminary underestimations of negative life events (b, -2.74; P =.003) and the difference between the number of optimistic and pessimistic predictions (b, 2.02; P =.003).

Overall the bias score was a significant predictor of euthymia time (weighted estimate, 0.47; 95% CI, 0.34-0.60). Next polarity (b, 0.242; P =.233) was not a significant predictor, indicating that the bias score was robust to either manic or depressive episodes.

Significant differences of bias scores were observed for vividness (t[34], 3.96; P <.001), emotional arousal (t[34], 3.971; P <.001), familiarity (t[34], 3.38; P =.002), prior experience (t[34], 3.304; P =.002), and negativity (t[34], 2.414; P =.021). Bias scores remained significantly associated with time in euthymia after controlling for these significant predictors (b, 0.351; P =.047).

This study was limited by its low sample size and by the fact that most patients (83.3%) were taking mood stabilizers. It remains unclear whether these medications had a significant impact on euthymia time prediction.

The study authors concluded that the time in euthymia was predicted by the belief updating after learning about actual probabilities of experiencing a negative life event. Calculating this bias score may be an important tool for practitioners to incorporate in their care for patients with bipolar disorder.


Ossola P, Garrett N, Sharot T, Marchesi C. Belief updating in bipolar disorder predicts time of recurrence. elife. 2020;9:e58891.

This article originally appeared on Psychiatry Advisor